Proc. Natl. Acad. Sci. USA Vol. 94, pp. 6330–6334, June 1997 Immunology

Specific complex formation between the type II -associated transactivators CIITA and RFX5

THOMAS SCHOLL*†,SANJEEV K. MAHANTA*‡, AND JACK L. STROMINGER§

Department of Molecular and Cellular Biology, Harvard University, 7 Divinity Avenue, Cambridge, MA 02138

Contributed by Jack L. Strominger, December 20, 1997

ABSTRACT Two of the defective in the five comple- sage for CIITA correlates with class II MHC expression in cell mentation groups identified in the class II-negative bare lines (8). Furthermore, this correlation extends to interferon ␥ lymphocyte syndrome or corresponding laboratory mutants (IFN-␥)-inducible expression and developmental extinction in have been cloned. One encodes a , RFX5, that is plasma cells since in both cases induced expression of CIITA a member of the RFX family of DNA binding . The alone causes class II MHC expression (11, 12). Thus, these results other, CIITA, encodes a large protein with a defined acidic strongly suggest that while several factors are re- transcriptional activation domain; this protein does not in- quired for class II MHC expression, CIITA is the specific teract with DNA. Expression plasmids encoding regions of regulator that controls both developmental and inducible class II RFX5 fused to the GAL4 DNA binding domain activated MHC . transcription from a reporter construct containing GAL4 CIITA does not bind DNA but does contain two functional sites in a cotransfection assay in the Raji human B cell line. domains. The N terminus encodes an acidic transcriptional However, these plasmids produced transcriptional activity in activator, and the C terminus targets the protein to MHC class HeLa cells only in conjunction with interferon ␥ stimulation, II promoters (13). The acidic activator is not specific for class a condition in which expression of both CIITA and class II II transcription and the function of this domain can be replaced major histocompatibility complex surface proteins are in- with an activation domain from a viral protein (13, 14). duced. Furthermore, these plasmids were not active in The gene for RFX5, defective in complementation group B, RJ2.2.5, an in vitro mutagenized derivative of Raji in which was cloned by the same approach as CIITA (15). RFX5 shares both copies of CIITA are defective. Transcriptional activation homology with a family of DNA binding proteins (RFX1–4), by the RFX5 fusion protein could be restored in RJ2.2.5 by a member of which was initially misidentified as the factor cotransfection with a CIITA expression plasmid. Finally, a responsible for the BLS defect (16) but was subsequently direct interaction between RFX5 and CIITA was detected with reported to regulate hepatitis B virus enhancer activity (17, the yeast two-hybrid and far-Western blot assays. Thus, RFX5 18). Two B cell lines derived from BLS patients in comple- can activate transcription only in cooperation with CIITA. mentation group B have defective RFX5 loci on both chro- RFX5 and CIITA associate to form a complex capable of mosomes, and class II MHC expression can be rescued in these activating transcription from class II major histocompatibil- lines by transfection of an expression plasmid encoding wild- ity complex promoters. In this complex, promoter specificity type RFX5 (15). The role of RFX5 in class II transcription is is determined by the DNA binding domain of RFX5 and the also supported by its specific binding to a conserved class II general transcription apparatus is recruited by the acidic promoter element, the X box, and its presence in nuclear activation domain of CIITA. complexes which form on the DRA promoter in electro- phoretic mobility shift assay (15). Type II combined immunodeficiency or bare lymphocyte syn- Herein, deletion mutagenesis demonstrates the regions of drome (BLS) is a rare congenital disease characterized by a lack RFX5 that are required for activating transcription in a B cell of transcription of major histocompatibility complex (MHC) class line. However, in HeLa cells this activity is detected only after II genes (1). BLS patients fall into three genetic complementation IFN-␥ stimulation. Furthermore, in a cell that is defective in groups and an in vitro mutant defined a fourth (2–5). Recently, a CIITA, this region of RFX5 is only functional after genetic potential fifth complementation group has been reported. The complementation with a CIITA expression plasmid. Finally, a atypical phenotype of these latter patients exhibited discoordi- direct interaction between RFX5 and CIITA was detected in nate regulation between ␣ and ␤ chains and low levels of class II vivo with the yeast two-hybrid assay and in vitro by direct surface expression on some cell types (6, 7). Transacting factors protein–protein interaction with a far-Western blot analysis. are responsible for these various defects since fusion of B cell lines Thus, RFX5 activates transcription indirectly by recruitment of derived from patients with those from a patient representing a CIITA specifically to the class II MHC promoter where the different complementation group or with a normal cell line previously characterized N-terminal acidic domain of CIITA restores class II expression. The defective genes in two of the serves to activate transcription. complementation groups have been cloned. Class II MHC trans- activator (CIITA) is defective in patients from complementation group A (8). Transfection experiments with plasmids expressing MATERIALS AND METHODS wild-type CIITA rescued MHC class II expression in three cell Isolation of CIITA and RFX5 cDNAs and Construction of lines from this group (8). Furthermore, cell lines from this group Plasmids. Raji cDNA was prepared by commercial mRNA carry mutations in both copies of the CIITA gene. Unlike other isolation (Oligotex, Qiagen, Chatsworth, CA) and synthesis factors reported to regulate class II MHC transcription for which genes have been cloned and which have all been shown to be Abbreviations: IFN-␥, interferon ␥; MHC, major histocompatibility ubiquitously transcribed DNA binding proteins (9, 10), the mes- complex; BLS, bare lymphocyte syndrome. *T.S. and S.K.M. contributed equally to this work. † The publication costs of this article were defrayed in part by page charge Present address: Myriad Genetic Laboratories, 320 Wakara Way, Salt Lake City, UT 84108. payment. This article must therefore be hereby marked ‘‘advertisement’’ in ‡Present address: Skirball Institute of Biomolecular Medicine, NYU accordance with 18 U.S.C. §1734 solely to indicate this fact. Medical Center, 540 First Avenue, New York, NY 10016. © 1997 by The National Academy of Sciences 0027-8424͞97͞946330-5$2.00͞0 §To whom reprint requests should be addressed.

6330 Downloaded by guest on September 30, 2021 Immunology: Scholl et al. Proc. Natl. Acad. Sci. USA 94 (1997) 6331

kits (Stratascript, Stratagene). This cDNA and oligonucleo- Yeast Two-Hybrid System and ␤-Galactosidase Assay. All tides (GIBCO͞BRL) for CIITA, CIITA.1 (ATAGAATTCG- yeast manipulations were performed in strain SFY526 in CAGCTGGAGAGGTCTCCAAC; contains EcoRI adapter accordance with the suppliers protocols (CLONTECH). All underlined) and CIITA.2 (ATACTCGAGTCTCAGGCT- quantitative data was derived from three independent colonies GATCCGTGAATC; contains XhoI adapter underlined); for transformed with the indicated plasmid(s). The ␤-galactosi- RFX5, RFX5.1 (GTGGCTACGAGTTTTCCAGATT) and dase assay was performed as described (24). RFX5.2 (GGAATACAGGTAAGGTCACTAC) were em- Far-Western Blot Analysis. Recombinant proteins from ployed in the PCR using pfu DNA polymerase (Stratagene) to plasmids pHISCIITTA1–161, pHISCIITA299–1129, and isolate the CIITA and RFX5 coding regions. These fragments pFUSE were induced in host Bl21(DE3) (Novagen) and were cloned into the EcoRI and XhoI sites of pBluescript II purified by nickel-chelate chromatography under denaturing KSϩ for CIITA and designated pBSCIITA or cloned into the conditions for CIITA-(299–1129) and native conditions for SmaI site for RFX5 and designated pBSRFX5. For mamma- CIITA-(1–161) and GAL4-(1–147) (Qiagen, Chatsworth, CA; lian expression and the yeast two-hybrid system, the CIITA protocols 7 and 5, respectively). Western blots were made and coding region was inserted into pBKCMV (Stratagene) from probed as described (24) except that 5 ␮g of recombinant which the lac promoter had been deleted and designated protein was loaded per lane and protein probes were 35S- pCMVCIITA and into pGAD424 (CLONTECH) in frame radiolabeled by in vitro translation (TNT reticulocyte, Pro- with the Gal4 activation domain and designated pGADCIITA. mega). Deletions of the RFX5 coding sequence were cloned in frame with the GAL4 DNA binding domain of pBXG1 using RESULTS restriction sites KpnI (bp 273) for essentially full-length pGAL- RFX⌬34, XhoI (bp 750) and PstI (bp 1125) without the DNA RFX5 Activates Transcription When Tethered to a Heter- binding domain for pGALRFX⌬194 and pGALRFX⌬318, ologous Promoter in Raji B Cells. A series of expression and EcoRI (bp 1600) without the proline-rich region for plasmids were constructed that fused coding regions of RFX5 pGALRFX⌬477. The RFX5 deletions resulting from digestion to the GAL4 DNA binding domain. The plasmids were with KpnI, XhoI, and PstI were also inserted into the yeast designed to remove from the 5Ј end successive putative two-hybrid expression vector pGBT9 and designated pG- functional domains identified within the RFX5 primary struc- BRFX⌬34, pGBRFX⌬194, and pGBRFX⌬318. For in vitro ture including the DNA binding and proline-rich domains and translation the XhoI and PstI derivatives of RFX5 were the central region of the coding sequence. These plasmids and inserted into pCITE2a (Novagen) and designated pTLN- the controls, pBXG1, which expresses the GAL4 DNA binding RFX5⌬194 and pTLNRFX5⌬318. For bacterial expression, domain, and pBXGAL-VP, which expresses the DNA binding CIITA regions corresponding to residues 1–161 and 299–1129 domain fused to the VP16 acidic activator, were cotransfected were cloned into pET15b (Novagen) and designated pHISCI- into Raji B cells with the reporter plasmid pG5EC, which ITTA1–161 and pHISCIITA299–1129. The plasmid pFUSE contains the GAL4 binding sites upstream of the chloram- contains the GAL4 DNA binding domain (residues 1–147) in phenicol acetyltransferase gene (25). Transcription was acti- pET15b. vated by the expression of two of these RFX5 fusion proteins All PCR products were completely sequenced, as were all corresponding to 5Ј deletions at amino acids residues 34 and junctions in cloning operations (Sequenase, United States 194 in the open reading frame (Fig. 1). These results demon- Biochemical). Plasmids pG5EC, pBXG1, pBXGAL-VP, and strate that portions of the RFX5 protein can activate tran- pFUSE were the generous gift of Mark Ptashne. Plasmids scription within B cells in a context other than that of the class pTD-1 and pVa3 were purchased (CLONTECH). II MHC promoter and that RFX5 can activate transcription Cell Culture and Transfection. Raji (CCL 86, American from the reporter construct in the absence of other MHC class Type Culture Collection) is a Burkitt lymphoma B cell line that II promoter binding proteins. Since activity was maintained expresses all MHC class II isotypes. RJ2.2.5 was derived by after deletion of the first 194 amino acids of RFX5, its DNA ␥-irradiation mutagenesis of Raji and is defective for global binding region was not required for transcriptional activation MHC class II transcription (19). The HeLa cell line (CCL 2, and could be substituted by the GAL4 DNA binding domain. American Type Culture Collection) was derived from a cer- Furthermore, comparison of the activities induced by the 34- vical carcinoma. Cells were cultured and treated with IFN-␥ as and 194-residue deletion constructs indicates that stronger reported (20). transcription activation was seen with the 194-residue deletion. Cells were transfected with the indicated amounts of plas- The presence of the RFX5 DNA binding domain may seques- mids by electroporation. Briefly, cells were washed twice in ter some of the fusion protein to recognition sites within the PBS with 10 mM Hepes (pH 7.3) and resuspended at 4 ϫ 107 genome and away from the reporter plasmid. cells per ml, on ice. Aliquots of 1 ϫ 107 cells were mixed with Transcriptional Activation by the RFX5 Activation Domain DNA, pulse-labeled at 250 V and 960 ␮Fd, maintained on ice Is IFN-␥-Inducible in HeLa Cells. HeLa cells are well char- for 10 min, and plated in 100-mm dishes with 15 ml of medium. acterized as a model system for IFN-␥ induction of class II After 48 hr of culture, chloramphenicol acetyltransferase MHC genes (26). When the RFX5 deletion plasmids were activities were determined as reported (21). Quantitative transfected into HeLa cells, no transcription from the reporter measurements of different chloramphenicol species using du- gene plasmid was detected. However, after treatment with plicate independent transfections of each sample were made IFN-␥, transcriptional activity in HeLa cells was observed in with phosphoimaging equipment (Fuji). As an internal control transfections with both the 34- and 194-residue deletion con- of transfection efficiencies 2 ␮g of pCMV␤ (22) was included structs (Fig. 2). Also, consistent with the data obtained from in all transfections. ␤-Galactosidase activity was measured the Raji cell transfections RFX5-(194–615) produced more spectrophotometrically. This result showed that transfection reporter gene activity than did RFX5-(34–615). These data efficiency generally varied less than 10% and always less than suggest that RFX5 does not contain an ‘‘activation domain’’ 15%; therefore, no normalization to this control was done. since previously defined activation domains are constituitively Southern Blot Analysis. CIITA mRNA was assessed in functional. Thus, either the RFX5 domain is modified in the Southern blots of PCR products from cDNA (prepared as cell after IFN-␥ treatment or, alternatively, an additional above) using primers CIITA.2 and CIITA.3 (TTGAGCGA- factor, induced or activated by IFN-␥ stimulation, might be CACGGTGGCGCTGTGGGAG). Blots were prepared, hy- required for transcriptinal activation by the RFX5 fusion bridized, and washed under high-stringency conditions as protein. Relevantly, IFN-␥ is known to induce the transcription described (23). of CIITA, a factor that is essential for transcription from class Downloaded by guest on September 30, 2021 6332 Immunology: Scholl et al. Proc. Natl. Acad. Sci. USA 94 (1997)

FIG. 1. Regions of RFX5 that activate transcription in the Raji human B cell line. A series of 5Ј deletions of the RFX5 coding region were fused to the GAL4-(1–147) DNA binding domain. These plasmids were designed to sequentially remove putative functional domains detected in the primary structure of RFX5. RFX5-(34–615) is essentially full-length, RFX5-(194–615) removes the DNA-binding domain, RFX5-(318–615) removes the central region, and RFX5-(477–615) removes the proline-rich region. Numbers used are the numbers of amino acid residues deleted from the reported translation initiation codon. pBXGAL-VP expresses the GAL4 DNA binding domain fused to the acidic activation domain of VP16-(413–490). Cells were transfected with 10 ␮g of pG5EC alone as a control for background activity by this reporter construct. The autoradiograph presents data from independent duplicate transfections of acetylated chloramphenicol species separated by TLC used to measure reporter gene activity. II MHC promoters (11). In fact, induction of CIITA message the B cell lymphoma line Raji. This line is globally defective in could be detected under the IFN-␥ stimulation conditions used MHC class II transcription due to mutations in both copies of in these assays (data not shown; see ref. 11). CIITA and serves as a CIITA negative background. Transcrip- CIITA Is Required for Transcriptional Activation by RFX5. tional activation by RFX5-(34–615) and RFX5-(194–615) was The cell line RJ2.2.5 was derived by laboratory mutagenesis of detected in transfection of RJ2.2.5 only when an expression plasmid encoding CIITA was included and not in controls in the absence of CIITA (Fig. 3). Thus, CIITA is required for the activity directed by RFX5. Direct Protein–Protein Interaction Between CIITA and RFX5. The requirement for CIITA in transcriptional activa- tion mediated through RFX5 could be explained by a direct interaction between these two factors. Two systems were employed to demonstrate this interaction. (i) The yeast two- hybrid system was employed (27). Three of the GAL4 RFX5 inserts in pBXG1 [RFX5-(34–615), -(194–615), and -(318– 615)] were cloned into the yeast expression plasmid pGBT9, and the full-length CIITA coding region was inserted into pGAD424. After transformation into yeast, an interaction was detected between CIITA and the fusion protein encoding residues 194–615 of RFX5 (Fig. 4). Unlike the mammalian cotransfection experiments, no activity was detected for the largest RFX5 region corresponding to amino acids 34–615. Furthermore, when yeast were transformed with the pGBT9 derivatives alone, no ␤-galactosidase activity was detected, confirming that RFX5 cannot activate transcription alone. (ii) Interaction between CIITA and RFX5 was also detected in vitro by the Far-Western technique. Three recombinant pro- FIG. 2. Transcriptional activity by RFX5–GAL4 fusion proteins is teins that included the N terminus of CIITA (residues 16–147), induced by IFN-␥ in HeLa cells. Autoradiographs of reporter gene activity from transfected control HeLa cells and HeLa cells treated the C terminus of CIITA (residues 299-1129), and the GAL4 with IFN-␥. Cells were transfected with 10 ␮g of the indicated DNA binding domain (residues 1–147) were expressed in plasmids. Samples transfected with GAL4 fusion proteins were co- bacteria in a vector that encoded an N-terminal histidine transfected with 10 ␮g of pG5EC. affinity tag. The proteins were then purified by nickel-chelate Downloaded by guest on September 30, 2021 Immunology: Scholl et al. Proc. Natl. Acad. Sci. USA 94 (1997) 6333

FIG. 3. Transcriptional activity by RFX4–GAL4 fusion proteins in RJ2.2.5 cells required CIITA. Autoradiographs of reporter gene activity from RJ2.2.5 cells cotransfected with the indicated plasmids and 10 ␮g of pG5EC and cotransfections that also included 10 ␮g of pCMVCIITA.

chromatography, refolded on replicate membranes, and quent data from this system when transfected into other cell probed with 35S-labeled in vitro-translated RFX5 regions types indicated first that this region of RFX5 did not function RFX5-(194–615) and RFX5-(318–615). As a positive control in all cell types and then that cooperation͞interaction with human TAF70 was included as a probe since it interacts with CIITA was required for this activity. Finally, a direct interac- the N-terminal acidic activation domain of CIITA (28). An tion was detected between RFX5 and CIITA both in vivo and interaction in the test samples was detected between the in vitro with the yeast two-hybrid assay and biochemically by C-terminal region of CIITA, residues 299-1129, and RFX5, far-Western blot analysis. residues 194–615, as well as between the N-terminal region of Thus, these results indicate that RFX5 and CIITA can form CIITA, residues 1–161, and TAF70 (Fig. 5). a complex that is capable of activating transcription. CIITA may function as a , binding together through pro- DISCUSSION tein–protein interactions RFX5 and other class II MHC pro- moter bound DNA binding proteins including at least those The isolation of genes encoding transcription factors respon- bound to the TATA box. Other interactions may influence the sible for the regulation of MHC class II genes by expression association of CIITA with class II MHC promoters. First, there and complementation screening techniques permits experi- are multiple cis elements that when mutated reduce transcrip- ments designed to elucidate the mechanisms through which tion from class II MHC promoters, including the X and Y box, these proteins function. Initial results from a functional assay the J and S elements, and an octamer element (29, 30). Unless using GAL4 fusion proteins suggested that a transcriptional these recognition sites bind factors whose expression specifi- ‘‘activation domain’’ was present in RFX5. However, subse- cally correlates with class II transcription, it is likely that they also ultimately function through CIITA. Additionally, coop- erative binding of multiple factors to class II promoters has been demonstrated. In fact, CIITA may tie together factors that bind to several DNA elements. In the case of the DRA promoter, the most strongly transcribed of the class II pro- moters, moreover, an additional proximal element, OCT,

FIG. 4. Interaction between RFX5 and CIITA detected in yeast. Histograms depict ␤-galactosidase reporter gene activity in yeast extracts from cells transformed with RFX5–GAL4 fusion protein expression plasmids or cotransformation that included a plasmid for expression of CIITA fused to the GAL4 activation domain. All data represent the mean activity detected in triplicate transformants. Individual measurements were within 5% of each other. pGBT9 encodes the GAL4 DNA binding domain alone. pTD-1 used in the positive control encodes a fusion protein made up of amino acids FIG. 5. Detection of an interaction between RFX5 and CIITA by 84–708 of simian virus 40 large tumor antigen and GAL4 activation far-Western blot analysis. Recombinant proteins (5 ␮g) were electro- domain and pVa3 encodes a fusion protein made up of GAL4 DNA phoresed and blotted on quadruplicate gels. Proteins were refolded on binding domain and amino acids 72–350 of p53 with which the simian the membranes and probed with the indicated radiolabeled in vitro- virus 40T antigen interacts. translated proteins. Downloaded by guest on September 30, 2021 6334 Immunology: Scholl et al. Proc. Natl. Acad. Sci. USA 94 (1997)

recruits the protein OCT2 and, through OCT2, BOB1, which 9. Liou, H.-C., Boothby, M. R., Finn, P. W., Davidon, R., Nabavi, interacts with the TATA box binding protein complex. Synergy N., Zeleznik-Le, N. J., Ting, J. P.-Y. & Glimcher, L. H. (1990) in binding may stabilize interactions with RFX5, CIITA, or Science 247, 1581–1584. both (31, 32). In these experiments, the levels of transcription 10. Sugawara, M., Scholl, T., Ponath, P. D. & Strominger, J. L. (1994) generated by the interaction of the RFX5–GAL4 fusion Mol. Cell. Biol. 14:8438–8450. 11. Steimle, V., Siegrist, C. A., Mottet, O., Lisowska-Grospierre, B. proteins was much less than that initiated from the DPA & Mach, B. (1994) Science 265, 106–109. proximal promoter (Fig. 2) or that detected when the CIITA 12. Silacci, P., Mottet, A., Steimle, V., Reith, W. & Mach, B. (1994) activation domain was fused directly to the GAL4 DNA J. Exp. Med. 180, 1329–1336. binding region. This finding may be a further indication that 13. Riley, J. L., Westerhide, S. D., Price, J. A., Brown, J. A. & Boss, additional interaction(s) to those with RFX5 are required for J. M. (1995) Immunity 2, 533–543. full recruitment of CIITA to or subsequent activation of the 14. Zhou, H. & Glimcher, L. H. (1995) Immunity 2, 545–553. MHC promoter. 15. Steimle, V., Durand, B., Barras, E., Zufferey, M., Hadam, M. R., CIITA is present in higher-order multiprotein complexes Mach, B. & Reith, W. (1995) Genes Dev. 9, 1021–1032. 16. Reith, W., Barras, E., Satola, S., Kobr, M., Reinhart, C., Herrero present on MHC class II promoters (13), and although it is Sanchez, C. & Mach, B. (1989) Proc. Natl. Acad. Sci. USA 86, required for MHC class II transcription and encodes a strong 4200–4204. acidic activation domain, it does not bind DNA. Conversely, 17. Reith, W., Ucla, C., Barras, E., Gaud, A., Durand, B., Herrero- RFX5 encodes a DNA binding domain that recognizes a DNA Sanchez, C., Kobr, M. & Mach, B. (1994) Mol. Cell. Biol. 14, element unique to MHC class II genes, the X box. Thus, CIITA 1230–1244. functions as a coactivator through specific protein interactions 18. Siegrist, C. A., Durand, B., Emery, P., David, E., Hearing, P., with RFX5. The role for RFX5 in this model is to lend Mach, B. & Reith, W. (1993) Mol. Cell. Biol. 13, 6375–6384. promoter specificity to this complex through X-box-specific 19. Accolla, R. S. (1983) J. Exp. Med. 157, 1053–1058. recognition by its DNA binding domain. The role of CIITA 20. Scholl, T., Stevens, M. B., Mahanta, S. & Strominger, J. L. (1996) J. Immunol. 156, 1448–1457. then is to activate transcription via its N-terminal acidic 21. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol. Cell. domain. Biol. 2, 1044–1051. 22. MacGregor, G. R. & Caskey, C. T. (1989) Nucleic Acids Res. 17, This work was supported by National Institutes of Health Grant 2365. CA47554 and by fellowships of the American Cancer Society to T.S. 23. Scholl, T., Pitcock, A. & Jones, B. (1992) Immunogenetics 36, and of the Irvington Foundation for Medical Research to S.K.M. 255–263. 24. Wu, J. Y. & Maniatis, T. (1993) Cell 75, 1061–1070. 1. Mach, B., Steimle, V. & Reith, W. (1994) Immunol. Rev. 138, 25. Sadowski, I. & Ptashne, M. (1989) Nucleic Acids Res. 17, 7539– 207–221. 7540. 2. Benichou, B. & Strominger, J. L. (1991) Proc. Natl. Acad. Sci. 88, 26. Boss, J. M. & Strominger, J. L. (1986) Proc. Natl. Acad. Sci. USA 4285–4288. 83, 9139–9143. 3. Hume, C. R. & Lee, J. S. (1989) Hum. Immunol. 26, 288–309. 27. Fields, S. & Song, O. (1989) Nature (London) 340, 245–246. 4. Hume, C. R., Shookster, L. A., Collins, N., O’Reilly, R. & Lee, 28. Sanjeev, K. Mahanta, Scholl, T., Yang, F.-C. & Strominger, J. L. J. S. (1989) Hum. Immunol. 25, 1–11. (1997) Proc. Natl. Acad. Sci. USA 94, 6324–6329. 5. Yang, Z., Acolla, R. S., Pious, D., Zegers, B. J. M. & Strominger, 29. Glimcher, L. H. & Cara, C. J. (1992) Annu. Rev. Immunol. 10, J. L. (1988) EMBO J. 7, 1965–1972. 13–50. 6. Hauber, I., Gulle, H., Wolf, H. M., Maris, M., Eggenbauer, H. & 30. Sugawara, M., Scholl, T., Ponath, P. & Strominger, J. L. (1994) Eibl, M. M. (1995) J. Exp. Med. 181, 1411–1423. Mol. Cell. Biol. 14, 8438–8450. 7. Douhan, J., Hauber, I., Eibl, M. M. & Glimcher, L. H. (1996) J. 31. Durand, B., Kobr, M., Reith, W. & Mach, B. (1994) Mol. Cell. Exp. Med. 183, 1063–1069. Biol. 14, 6839–6847. 8. Steimle, V., Otten, L. A., Zufferey, M. & Mach, B. (1993) Cell 75, 32. Moreno, C. S., Emery, P., West, J. E., Durand, B., Reith, W., 135–146. Mach, B. & Boss, J. (1995) J. Immunol. 155, 4313–4321. Downloaded by guest on September 30, 2021